Abstract for talk at USC, 1/26/01:

Effective and efficient electronic commerce requires that our machines "talk" to each other and be able to "say" the kinds of things people say when conducting business. Electronic Data Interchange (EDI) standards—X12, EDIFACT, S.W.I.F.T., &c.— were early forays in designing and fielding machine-understandable languages for such purposes. Although the object of much promotion, EDI's scope of economic application remains much less than has been widely hoped. This has limited the degree to which business transactions can be fully automated, and hence the degree to which costs can be reduced and services improved. Newer protocols, based on XML and using the Internet for transmission, are now being promoted as replacements for the old EDI standards, but the economic benefits have not proved compelling.

In this talk I review the prospects for, and present new results pertaining to, development of general-purpose formal languages for business communication (FLBC). A most salient fact is that business communications—including such "simple" documents as purchase orders, bills of lading, receiving reports, &c.—are semantically very complex. EDI standards, and now XML protocols, have lacked workable semantic theories and have consequently taken ad hoc approaches in their designs. This is a fundamental reason why their scope of economically attractive application has proved to be so limited. The talk will focus on presenting a positive theory—the author's lean events semantics—that can overcome these difficulties.